1

CHAPTER 5
COLLISIONS

5.1 Introduction

In this chapter on collisions, we shall have occasion to distinguish between elastic and inelastic
collisions. An elastic collision is one in which there is no loss of translational kinetic energy. That
is, not only must no translational kinetic energy be degraded into heat, but none of it may be
converted to vibrational or rotational kinetic energy. It is well known, for example, that if a ball
makes a glancing (i.e. not head-on) elastic collision with another ball of the same mass, initially
stationary, then after collision the two balls will move off at right angles to reach other. But this is
so only if the balls are smooth. If they are rough, after collision the balls will be spinning, so this
result – and any other results that assume no loss of translational kinetic energy − will not be valid.
When molecules collide, they may be set into rotational and vibrational motion, and in that case the
collision will not be elastic in the sense in which we are using the term. If two atoms collide, one
(or both) may be raised to an excited electronic level. Some of the translational kinetic energy has
then been converted to potential energy. If the excited atom subsequently drops down to a lower
level, that energy is radiated away and lost from the system. Superelastic collisions are also
possible. If one atom, before collision, is in an excited electronic state, on collision it may make a
radiationless downwards transition, and the potential energy released is then converted to
translational kinetic energy, so the collision is superelastic. None of this is intended to mean that
elastic collisions are impossible or even rare. In the case of collisions involving macroscopic
bodies, such as smooth, hard billiard balls, collisions may not be 100% elastic, but they may be
close to it. In the case of low-energy (low temperature) collisions between atoms, there need be no
excitation to excited levels, in which case the collision will be elastic. Some subatomic particles, in
particular leptons (of which the electron is the best-known example), are believed to have no
internal degrees of freedom, and therefore collisions between them are necessarily elastic.

In laying out the principles involved in collisions between particles, we need not suppose that the
particles actually "bang into" – i.e. touch – each other. For example most of the principles that we
shall be describing apply equally to collisions between balls that "bang into" each other and to
phenomena such as Rutherford scattering, in which an alpha particle is deviated from its path by a
gold nucleus without actually "touching" it. Of course, if you think about it at an atomic level,
when two billiard balls collide, the atoms don't actually "touch" each other; they are repelled from
each other by electromagnetic forces, just as the alpha particle and the gold nucleus repelled each
other in the Rutherford-Geiger-Marsden experiment.

The theory of collisions is used a great deal, of course, in the study of high-energy collisions
between particles in particle physics. Bear in mind, however, that in "atom-smashing" experiments
with modern huge particle accelerators, or even in relatively mild collisions such as Compton
scattering of x-rays, the particles involved are moving at speeds that are not negligible compared
with the speed of light, and therefore relativistic mechanics is needed for a proper analysis. In this
chapter, collisions are treated entirely from a nonrelativistic point of view.
 2
5.2 Bouncing Balls

When a ball is dropped to the ground, one of four things may happen:

1. It may rebound with exactly the same speed as the speed at which it hit the ground. This is an
elastic collision.

2. It may come to a complete rest, for example if it were a ball of soft putty. I shall call this a
completely inelastic collision.

3. It may bounce back, but with a reduced speed. For want of a better term I shall refer to this as
a somewhat inelastic collision.

4. If there happens to be a little heap of gunpowder lying on the table where the ball hits it, it may
bounce back with a faster speed than it had immediately before collision. That would be a
superelastic collision.

speed after collision

The ratio is called the coefficient of restitution, for which I shall use the
speed before collision
symbol e. The coefficient is 1 for an elastic collision, less than 1 for an inelastic collision, zero for
a completely inelastic collision, and greater than 1 for a superelastic collision. The ratio of kinetic
energy (after) to kinetic energy (before) is evidently, in this situation, e2.

If a ball falls on to a table from a height h0, it will take a time t 0 = 2 h0 / g to fall. If the collision is
somewhat inelastic it will then rise to a height h1 = e 2 h0 and it will take a time et to reach height h1.
Then it will fall again, and bounce again, this time to a lesser height. And, if the coefficient of
restitution remains the same, it will continue to do this for an infinite number of bounces. After a
billion bounces, there is still an infinite number of bounces yet to come. The total distance
travelled is

h = h0 + 2h0 (e 2 + e 4 + e 6 + K) 5.2.1

and the time taken is t = t0 + 2t0 ( e + e 2 + e3 + K). 5.2.2

These are geometric series, and their sums are

 1 + e2 
h = h0  ,
2 
5.2.3
1 − e 

which is independent of g (i.e. of the planet on which this experiment is performed), and

1 + e
t = t0  . 5.2.4
1 − e 
 3

For example, suppose h0 = 1 m, e = 0.5, g = 9.8 m s−2, then the ball comes to rest in 1.36 s after
having travelled 1.67 m after an infinite number of bounces. Discuss. (E.g. Does it ever stop
bouncing, given that, after every bounce, there is still an infinite number yet to come; yet after 1.36
seconds it is no longer bouncing...?)

5.3 Head-on Collision of a Moving Sphere with an Initially Stationary Sphere

Before: m1 u m2

After: m1 v1 m2 v2

FIGURE V.1
The coefficient of restitution is

relative speed of recession after collision .

e= 5.3.1
relative speed of approach before collision

We suppose that the two masses m1 and m2, the initial speed u, and the coefficient of restitution e
are known; we wish to find v1 and v2. We evidently need two equations. Since there are no
external forces on the system, the linear momentum of the system is conserved:

m1u = m1v 1 + m2v 2 . 5.3.2

The second equation will be the restitution equation (equation 5.3.1):

v 2 − v 1 = eu. 5.3.3

These two equations can be solved to yield

 m − m2 e 
v1 =  1  u 5.3.4
 m1 + m2 
 4
 m (1 + e) 
and v 2 =  1  u . 5.3.5
 m1 + m2 

The relation between the kinetic energy loss and the coefficient of restitution isn't quite as simple as
in section 5.2. Exercise. Show that

kinetic energy (after) mv2 + mv2 m + m2 e 2 .

= 1 1 2 1 2 = 1 5.3.6
kinetic energy (before) m1u m1 + m2

If m2 = ∞ (as in section 5.2), this becomes just e2. If e = 1, it becomes unity, so all is well.

If m1<< m2 (Ping-pong ball collides with cannon ball), v1 = −u , v2 = 0.

If m1 = m2 (Ping-pong ball collides with ping-pong ball), v1 = 0 , v2 = u.

If m1 >> m2 (Cannon ball collides with ping-pong ball), v1 = u, v2 = 2u.

Example: A moving sphere has a head-on elastic collision with an initially stationary sphere.
After collision the kinetic energies of the two spheres are equal. Show that the mass ratio of the
two spheres is 0.1716. Which of the two spheres is the more massive? (I guarantee that your
answer to this will be correct.)

5.4 Oblique Collisions

In figure V.2 I show two balls just before collision, and just after collision. The horizontal line is
the line joining the centres – for short, the "line of centres". We suppose that we know the velocity
(speed and direction) of each ball before collision, and the coefficient of restitution. The direction
of motion is to be described by the angle that the velocity vector makes with the line of centres.
We want to find the velocities (speed and direction) of each ball after collision. That is, we want to
find four quantities, and therefore we need four equations. These equations are as follows.

u2
u1

Before: α1 α2 Line of centres

m1 m2

v1
v2

β1 β2 Line of centres
After:
m1 m2

FIGURE V.2
 5

There are no external forces on the system along the line of centres. Therefore the component of
momentum of the system along the line of centres is conserved:

m1v1 cos β1 + m2v 2 cos β 2 = m1u1 cos α1 + m2u 2 cos α 2. 5.4.1

If we assume that the balls are smooth -.i.e. that there are no forces perpendicular to the line of
centres and the balls are not set into rotation, then the component of the momentum of each ball
separately perpendicular to the line of centres is conserved:

v1 sin β1 = u1 sin α1 5.4.2

and v 2 sin β 2 = u 2 sin α 2 . 5.4.3

The last of the four equations is the restitution equation

relative speed of recession along the line of centres after collision .

e=
relative speed of approach along the line of centres before collision

That is, v 2 cos β 2 − v1 cos β1 = e(u1 cos α1 − u 2 cos α 2 ). 5.4.4

Example. Suppose m1 = 3 kg, m2 = 2 kg, u1 = 40 m s−1 u2 = 15 m s−1

α1 = 10ο , α2 = 70o , e = 0.8

Find v1 , v2 , β1 , β2 .

Answers: v1 = 16.28 m s−1 v2 = 44.43 m s−1

β1 = 25o 15' β2 = 18o 30'

Example. Suppose m1 = 2 kg, m2 = 3 kg, u1 = 12 m s−1 u2 = 15 m s−1

α1 = 20ο , α2 = 50o , β2 = 47o

Find v1 , v2 , β1 , e .

Answers: v1 = 10.50 m s−1 v2 = 15.71 m s−1

β1 = 23o 00' e = 0.6418

Problem. If u2 = 0, and if e = 1 and if m1 = m2 , show that β1 = 90o and β2 = 0o .

 6

5.5 Oblique (Glancing) Elastic Collisions, Alternative Treatment

m1 u V m2
Before: %
C v1
m1
θ
After: α1
FIGURE V.3 m2

v2

In figure V.3, unlike figure V.2, the horizontal line is not intended to represent the line of centres.
Rather, it is the direction of the initial velocity of m1, and m2 is initially at rest. The second mass m2
is slightly off the line of the velocity of m1. I am assuming that the collision is elastic, so that
e = 1. In the "before" part of the figure, I have indicated, as well as the two masses, the position
and velocity V of the centre of mass C. The velocity of C remains constant, because there are no
external forces on the system. I have not drawn C in the "after" part of the figure, because it would
get a little in the way. Think about where it is.

Figure V.3 shows the situation in "laboratory space". (Later, we'll look at the situation referred to a
reference frame in which C is at rest – "centre of mass space".) The angle θ is the angle through
which m1 has been scattered (the "scattering angle"). I have indicated in the figure how it is related
to the α1 and β1 of section 5.3. Note that m2 (initially stationary) scoots off along the line of
centres.

The following two equations express the constancy of linear momentum of the system.

(m1 + m2 )V = m1u = m1v1 cos θ + m2v 2 cos α1. 5.5.1

I'm going to draw, in figure V.4, the situation "close-up", so that you can see the geometry more
clearly. Note that the distance b is called the impact parameter. It is the distance by which the
two centres would have missed each other had the first particle not been scattered.

In figure V.5, I draw the situation in centre of mass space, in which the centre of mass C is
stationary. In this reference frame, I just have to subtract V from all the velocities. Note that in
centre of mass space the speeds of the particles are unaltered by the collision. In centre of mass
space, m1 is scattered through an angle θ', and I am going to find a relation between θ' , θ and the
mass ratio m2/m1.

I shall start with the profound statement that

 7

v 1 sin θ .
tan θ = 5.5.2
v 1 cos θ

Before After

v1
u θ
m1 α1 + θ = β1
α1 m1 α1
R1

b = ( R1 + R2 ) sin α1 →
R2

v2
m2
m2

FIGURE V.4

m1 u' = u − V V m2
Before: %
C
u'

m1

After: % θ'
m2

FIGURE V.5
 8

Now v1 sin θ is the y-component of the final velocity of m1 in laboratory space. The y-component
of the final velocity of m1 in centre of mass space is u ' sin θ' , and these two are equal, since the y-
component of the motion is unaffected by the change of reference frame. Therefore

u ' sin θ' .

tan θ = 5.5.3
v 1 cos θ

Therefore v1 sin θ = u ' sin θ' . 5.5.4

The x-components of the "before" and "after" velocities of m1 are related by

v1 cos θ = u ' cos θ' + V . 5.5.6

Substitute equations 5.5.4 and 5.5.6 into equation 5.5.2 to obtain

sin θ' .
tan θ = 5.5.7
cos θ' + V /u '

But (m1 +m2 )V = m1u = m1 (u '+V ), 5.5.8

V m
from which = 1. 5.5.9
u' m2

On substituting this into equation 5.5.7, we obtain the relation we sought:

sin θ ' .
tan θ = 5.5.10
cos θ' + m1 / m2

This relation is illustrated in figure V.6 for several mass ratios.

Let us try to interpret the figure. For m2 > m1, any scattering angle, forward or backward, is
possible, but for m2 < m1, backward scattering is not possible, and forward scattering is possible
only up to a maximum. This is only to be expected. Thus for an impact parameter of zero or of
R1 + R2 , and m2 < m1, the scattering angle θ must be zero, and therefore for intermediate impact
parameters it must go through a maximum. This would be clearer if we could plot the scattering
angle versus the impact parameter, and indeed that is something that we shall try to do. In the
meantime it is easy to show, by differentiation of equation 5.5.10 (do it!), that the maximum
scattering angle is sin −1 µ , where µ = m2 / m1 . That is, if the scattered particle is very massive
compared with the scattering particle, the maximum scattering angle is small – just to be expected.
 9

180

160
θ FIGURE V.6
140

120 µ = m2/m1 = 8 4

100 2 10/9

1
80
9/10
60

40
0.5
20 0.25
0.125
0
0 20 40 60 80 100 120 140 160 180
θ'

I want to do two things now - one, to calculate the scattering angle θ as a function of impact
parameter, and two, to calculate v1 / u as a function of scattering angle. I’m going to start with
equations 5.4.1,2 and 4, except for the following. I’ll assume e = 1 (elastic collision), and u2 = 0
(m2 is initially stationary), and β2 = 0 (since m2 is initially stationary, it must move along the line of
centres after collision). Since I want to try to calculate the scattering angle, I’ll write
θ + α1 for β1 (see figure V.4). I’m also going to write r1 , r2 and µ for the dimensionless ratios
v1/u, v2/u and m2/m1 respectively. With those small changes, equations 5.4.1,2 and 4 become

r1 cos(θ + α1 ) + µr2 = cos α1 , 5.5.11

r1 sin(θ + α1 ) = sin α1 , 5.5.12

r2 − r1 cos(θ + α1 ) = cos α1 . 5.5.13

Eliminate r2 from equations 5.5.11 and 5.5.13 to obtain

r1 cos(θ + α1 ) = M cos α1 , 5.5.14

 10

1−µ m − m2 .
where M = = 1 5.5.15
1+ µ m1 + m2

If we now eliminate α1 from equations 5.5.12 and 14, we obtain the relation between v1 / u and the
scattering angle, which was the second of our two aims above. The elimination is easily done as
follows. Expand sin and cos of θ + α1 in the two equations, divide both sides of each equation by
cos α1 and eliminate tan α1 between the two equations. The result is

r12 − r1 (1 + M ) cos θ + M = 0 . 5.5.16

We’ll have a look at this equation in a moment, but in the meantime, instead of eliminating α1 from
equations 5.5.12 and 14, let’s eliminate r1. This will give us a relation between the scattering angle
θ and α1, and, since α1 is closely relation to the impact parameter (see figure V.4) this will achieve
the first of our aims, namely to find the scattering angle as a function of the impact parameter. If
you do the algebra, you should find that the relation between θ and α1 is

a (1 − M ) ,
t = 5.5.17
a2 + M

where t = tan θ and a = tan α1 . 5.5.17a,b

b
Now let b' = 5.5.18
R1 + R2

and from figure V.5 we see that b' = sin α1 . 5.5.19

On elimination of α1 from equations 5.5.17 and 5.5.19, we obtain the required relation between
scattering angle θ and (dimensionless) impact parameter b':

2µb' 1 − b'2
tan θ = . 5.5.20
1 − µ + 2µb'2

This relation is shown in figure V.7. The values of the mass ratio µ ( = m2 / m1 ) are (from the
lowest up) 18 , 14 , 12 , 109 , 1 , 109 , 2 , 4 , 8 and (dashed) ∞. This figure is perhaps slightly easier to
interpret than figure V.6. One can see that for µ > 1, any scattering angle is possible, but for µ < 1,
the scattering angle has a maximum possible value, less than 90o, and the scattering angle is zero
for b' = 0 or 1.

Exercise. We saw, by differentiation of equation 5.5.10, that the maximum scattering angle was
sin −1 µ . Now show the same thing by differentiation of equation 5.5.20. (This is not so easy, is it?)
Show that the scattering angle is greatest for an impact parameter of
 11

1− µ .
b' = 5.5.21
2

You will notice that, for b' = 0 (head-on collision) the scattering angle changes abruptly from 0 to
180o as the mass ratio changes from less than 1 to more than 1. No problem there. But if the mass
ratio is exactly 1 (not the tiniest bit less or the tiniest bit more) the scattering angle is apparently
90o. This may cause some puzzlement until it is realized that for a head-on collision with µ = 1 the
first sphere comes to a dead halt.

The case of µ = ∞ (second sphere immovable) is of some interest. It is easy in that case to calculate
how the scattering angle varies with impact parameter for an elastic collision, merely by requiring
the scattered sphere to obey the law of reflection, and without any reference to equation 5.5.20.

180

160 FIGURE V.7

140

120
o
θ
100

80

60

40

20

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
b'

Easy Exercise. Without any reference to equation 5.5.20, show that, if the second ball is
immovable, the scattering angle is related to the impact parameter by

θ = 180o − 2 sin −1 b' . 5.5.22

 12
Not-so-easy exercise. Show that, in the limit as µ → ∞ , equation 5.5.20 approaches equation
5.5.22.

In any case, the limiting case as the second sphere becomes immovable is shown as a dashed curve
in figure V.7.

Exercise of Intermediate Difficulty. The mass ratio m2/m1 is 0.9, and the scattering angle is 50o.
What was the impact parameter?
Answers b' = 0.07270 or 0.58540.

We have now dealt with the direction of motion of m1 after scattering as a function of impact
parameter. We should now look at the speed of m1 after collision, and this takes us back to
equation 5.5.16, which gives is the speed (r1 = v1 / u ) as a function of scattering angle θ. It is
quadratic in r1, so, for a given scattering angle there are two possible speeds – which is not
surprising, because a given scattering angle can arise from two different impact parameters, as we
have just found out. We can conveniently show the relation between r1 and θ simply by plotting
the equation in polar coordinates. I’ll re-write the equation here for easy reference:

r12 − r1 (1 + M ) cos θ + M = 0 . 5.5.16

1−µ m − m2 ,
Here, M = = 1 but I want to write the equation in terms of the mass fractions
1+ µ m1 + m2

m1 1 m2 µ .
q1 = = and q2 = = 5.5.23a,b
m1 + m2 1 + µ m1 + m2 1 + µ

If you work at this for a short while, you will find that equation 5.5.16 becomes

r12 + q12 − 2r1q1 cos θ = q22 , 5.5.24

and one is then overcome with an overwhelming desire to draw a triangle:

r1

q2

θ
q1
FIGURE V.8
 13

For a given mass ratio, the locus of r1 (the speed) versus θ (the scattering angle) is such that q1 and
q2 are constant – in other words, it is a circle:

r1

q2

θ
¢ q1
FIGURE V.9

v1

u θ
¢ 1 1
4 8
1
2
1
2
4
µ = 8

FIGURE V.10
 14

One can imagine that the first particle comes in from the left at speed u and the collision takes place
at the asterisk, and, after collision, it is moving at a speed r1 times u in a direction θ, the magnitude
of its velocity vector being determined by where the vector intersects the circle (in two possible
places) given by equation 5.5.24. The maximum scattering angle corresponds to a velocity vector
that is tangent to the circle. If the asterisk is the pole (origin) of the polar coordinates, the centre of
the circle is at a distance q1 from the pole, and its radius is q2. Figure V.10 shows the circles
corresponding to several mass ratios. The figure graphically illustrates the relation between u, v1 ,
θ and µ. You can see, for example, that if µ > 1, scattering through any angle is possible, and the
relation between v1 and θ is unique; but if µ < 1, only forward scattering is possible, up to a
maximum θ, and, for a given θ, there are two solutions for v1.

This deals with what happens to the sphere m1. We can now turn our attention to m2. Starting from
equations 5.5.11, 12 and 13, we are going to want to eliminate r1 and θ - indeed anything that
pertains to the sphere m1.

If you refer to figure V.4 you will see that, after collision, m2 scoots off at an angle α1 to the
original direction of motion of m1. Therefore I think it is of interest to find a relation between
r2 (= v 2 / u ) and α1. If we succeed in doing this, it means that we can also find a relation, if we
want it, between r2 and the impact parameter, since b' = sin α1. It is easy to eliminate r1 from
equations 5.5.12 and 13, and then you can get tan (θ + α1) from equation 5.5.14, and hence get the
required relation:

2 cos α1 .
r2 = 5.5.25
1+ µ

I’ll draw this relation as a polar graph, r2 versus α1, in figure V.11. I’ll leave the reader to work out
and draw the relation between r2 and b' if he or she wishes. Equation V.11 is the polar equation to
a circle of radius 1/(1 + µ).

Example. Suppose the mass ratio µ = m2 / m1 = 0.5 and the scattering angle is θ = 20o. Equation
5.5.16 or figure V.10 will show that r1 = 0.8696 or 0.3833. Equation 5.5.17 will show that
α1 = 58o.4 or 11o.6 . And equation 5.5.25 or figure V.11 will show that r2 = 0.6983 or 1.306.
I’ll leave it to the reader to determine which alternative values of r1 , r2 and α1 go together.
 15

1 1
u 8 4 2 1
¢
2 4
µ= 1
α1 8

v2

FIGURE V.11